Re: [ba-ohs-talk] Concept: Typed Versioning
On Mon, 2002-03-25 at 01:55, Eugene Eric Kim wrote:
> On 22 Mar 2002, Lee Iverson wrote:
>
> > It's still an open question as to what a "link type" really is. If we
> > stick to the data-knowledge-context modularization that I've argued for,
> > then "semantic" links could be held in the "data" system, but would need
> > some sort of external maintenance.
>
> Why?
>
> I understand how the external maintenance would work. But why store link
> semantics outside of the box in the first place? (01)
The issue as I'd frame it is more of what semantics can we profitably
build into the basic data infrastructure and what should we depend on
other tools to do. The issue as far as I see it is not about storage,
we should be able to *store* any kind of extensible link semantics into
the NODAL data model. The issue is what can the NODAL repository be
seen as responsible for maintaining. (02)
I'm happy to take on the task of maintaining some sort of stability
maintenance tasks *inside* NODAL, for example making sure that link
targets actually point to something when they are internal references.
I'd also be perfectly happy with having NODAL responsible for
maintaining a queryable backlink index (and perhaps even something more
powerful than others have imagined). More on that below. (03)
> If you don't store link semantics in the system, then you can't do things
> like represent the data model of a Topic Map. You could represent the
> data model of an XTM document, but those are two different things. (04)
As I said, since I'm happy storing *anything* in a NODAL repository, and
since I definitely agree that hyperlinks are fundamental, and that
attaching semantic markers to hyperlinks is a reasonable basic
requirement, I think we're on the same page here. When I refer to other
tools, I'm referring to the tools that either *understand* or *embody*
the semantics of those markers and can manipulate them *semantically*. (05)
[Newish idea alert]
It seems so basic, but I haven't seen anybody mention it before. If
anybody has a reference to a system that does this, I'd be keen to see
it. (06)
The HTTP header *may* contain a REFERER (note mispelling) reference,
which is actually evidence that a hyperlink on the REFERER page is
what was used to get to this requested page. Just keeping track of
this mapping is a great way to model the outside world's view of your
repository. A more substantial solution would involve reading those
referred pages in order to build up some representation of the context
of that referral and possibly other related pointers (kind of a
single-link take on the Google evaluation methodology). (07)
--
-------------------------------------------------------------------------------
Lee Iverson
leei@ece.ubc.ca Dept of ECE, 2356 Main Mall
http://www.ai.sri.com/~leei/ Vancouver BC Canada V6T 1Z4
Office: (604) 822-3381 (08)